Airborne reverse flow immersion thermometers were designed to prevent sensor wetting in cloud. Yet there is strong evidence that some wetting does occur and therefore also sensor evaporative cooling as the aircraft exits a cloud. Numerous penetrations of cumulus clouds in a broad range of environmental and cloud conditions are used to estimate the resulting negative temperature bias. This cloud exit "cold spike" can be found in all cumulus clouds, even at subfreezing temperatures, both in continental and maritime cumuli. The magnitude of the spike correlates most strongly with the dryness of the ambient air. A temperature correction based on this relationship is proposed. More important than the cloud exit cold spike, from a cumulus dynamics perspective, is the negative bias within cloud. Such bias is expected, due to evaporative cooling as well. Evaporation from the wetted sensor in cloud is surmised because air decelerates into the thermometer housing, and thus is heated and becomes subsaturated. Thus an in-cloud temperature correction is proposed, based on the composite cloud exit evaporative cooling behavior. This correction leads to higher and more realistic estimates of cumulus buoyancy and lower estimates of entrainment.